数据的海量存储需求对传统数据存储模式带来了巨大的挑战。目前大多数集群/分布式文件系统都被设计用来管理大文件。当在处理海量的小文件时,这些文件系统并不具有良好的性能表现。论文提出了一种面向海量小文件的元数据模型,该模型是基于具有超级节点的混合P2P拓扑结构,对小文件的优化管理具有一定的普适性。通过测试对比发现,在小文件读写性能上有数量级的提升。
The massive requirement of data storage brings a great challenge to traditional mode. Most cluster/distributed file systems are designed to manage large files and suffers a performance penalty while managing a large amount of small files. This paper proposes a mas sive small file metadata management model based on the features of a hybrid P2P structure with some backup super peers. And its optimiza tion of small files is universal. Finally, a prototype system for testing is implemented and a high performance in small file 1/O is obtained.